Belief updating on Planet Bayes

Readers of the New Yorker (which describes itself in Trumpian terms as the best magazine that ever was) were recently told the following:

One study suggests that two-thirds of Americans between the ages of fifteen and thirty-four who were treated in emergency rooms suffered from injuries inflicted by police and security guards, about as many people as the number of pedestrians injured by motor vehicles.

Louise Perry reviewed the scientific study on which is was based:

I did my best to work out a rough estimate of the true proportion of 15-34 year olds visiting the ER who had suffered legal intervention injuries, and arrived at a figure of 0.2% (you can follow my working in this thread). So I believe Lepore’s claim to be off by a factor of several hundred.

Because New Yorker readers are a superior breed, with Bayesian reasoning skills, they will dutifully reduce their estimate of the frequency of police beatings by a factor of 300, once they are informed of Ms. Perry’s correction.

Those that don’t see the correction will assume that 1/3 of all ER admissions of young people (at a minimum) are for people who were beaten by police after they were injured in a traffic accident, in what can only be described as adding insult to injury. That’s because New Yorker readers are good at math, and understand that 2/3 plus 2/3 equals 4/3.

PS. I learned early on not to trust the press. The photo below describes a dispute between my dad and a neighbor over a garage. I recall (back in the 1960s) my dad telling me that the newspaper got the facts wrong. And that’s despite the fact that my dad’s brother was a reporter for this paper (The Capital Times). I was reminded of this story by a Matt Yglesias tweet that I saw today, describing a very similar dispute in Maine.

Tags:

30 Responses to “Belief updating on Planet Bayes”

1. Benjamin Cole
21. July 2020 at 19:00

In my youth and until 20-30 years ago, I worshipped The New Yorker.

I gave up my sub long ago, and now try to read five articles free a month on their website. Even that is trying.

Yes, after getting injured in auto accidents, young Americans then combat police and end up in emergency rooms. Very typical.

However, the English language now flummoxes The New Yorker. Lepore may be hopping topics a bit.

Perhaps Lepore is saying that the number of pedestrians injured in auto accidents in the US, in all age groups, is equal to the number of young people who enter ER due to fighting with police and security guards. Her use of language is muddy.

That reminds of an old line from the now-departed Jack Smith, truly a talented writer for the old L.A. Times:

“I was on the porch chasing a cat with a broom in my underwear.”

You can read that a lot of ways.

That the New Yorker would sink into mire of woke “journalism” truly sad. There it joins the NYT and WaPo.

These used to be great news outlets, if given sometimes to excess and vanity.

2. Matthias Görgens
21. July 2020 at 19:16

I hope the Economist holds the line. Though I have become a bit disillusioned with them as well, but not in the manner described about the New Yorker here.

Scott, a bit off-topic: last Friday I read an article reporting on some noises from the Fed that they are considering overshooting their inflation target this time and in the future to make up for lapses. In essence, it sounded like an endorsement of average inflation targeting.

Have you heard of this? I would have expected that announcement to be bigger news here and on econlog?

3. Jg
21. July 2020 at 19:37

Correction – scott only trusts the press when it comes to quotes critical of Trump.

4. Jason
21. July 2020 at 21:31

Good for you Scott,
I was worried you had degenerated into just another boring anti Trumper. I dislike Trump immensely but the media will always hold him to account.

They will NEVER hold Democrats to account like Andrew Cuomo in his nursing home scandal. Conservatives are fettered, but Democrats always get a free pass.

And Trump is too lazy to pull an Emperor Palpatine.

5. Christian List
22. July 2020 at 01:39

So far Jill Lepore gave me the impression to be rather moderate (compared to her environment) and interested in facts, it is a pity (and telling) that such a mistake happens to her.

I don’t see the study itself but 1-2 press releases about it are a bit misleading in this regard. It’s not explicitly worded that that the data is already presorted, I assume under the assumption that no reasonable person with contact to people aged 15-34 would come up with the idea that 64% of the physical injuries in emergency rooms in this age group are caused by security authorities.

So in this respect it is very telling that Lepore simply accepts her assumption uncritically, apparently without thinking about it much longer. In her world view it is completely plausible that 64% of physical injuries in emergency rooms are caused by the police. These people do have contact to real life or in what kind of bubble do they live in?

They are the new Pope. Rule 1 we are infallible and in case we are wrong, then rule 1 automatically comes into effect.

6. Christian List
22. July 2020 at 01:44

Part of my text got deleted while typing:

I find it interesting that the essay has still not been corrected. Apparently they are still not sure whether it is really a mistake or not. They act like a new Pope…

7. foosion
22. July 2020 at 02:33

“understand that 2/3 plus 2/3 equals 4/3”

The New Yorker quote does not imply 2/3 plus 2/3. The 2/3 is a percentage of 15-34 year olds. Pedestrians injured is a number. The total population has many people who are not 15-34.

Assume of population of 100 people, 12 of whom are 15-34. Two-thirds of 15-34 is 8 people. It is possible that 8/88 were injured by a motor vehicle.

8. Tacticus
22. July 2020 at 04:08

It appears to me that whomever wrote the press-release for the study ( https://www.hsph.harvard.edu/news/hsph-in-the-news/injuries-caused-by-police-security-guards-rising/ ) made a terrible error and that the New Yorker only read the release, not the actual study.

The press release states that ‘Sixty-four percent of the estimated 683,033 injuries logged between 2001-2014 among persons age 15-34 resulted from an officer hitting a civilian.’

The study itself states that from 2001-2014, an estimated 683,033 men and women between the ages of 15-34 went to the ER for ‘legal intervention injuries.’ It says nothing about the total amount of ER visits, so the comment about ‘sixty-four percent’ or ‘two-thirds’ is coming out of nowhere.

Also: ‘For men in this age group, the rate of visits for such injuries (98.7 per 100,000) was on par with that of visits for pedestrians injured by motor vehicles (101.1 per 100,000).’ So, for men in that age group, they are about as likely to be injured in a legal intervention than the average American is to be hit by a car whilst walking.

The study itself is interesting, showing that the incidence of legal intervention injuries went up from 2001-2014, especially for men and especially for black persons, in case people doubted that.

9. Alan Goldhammer
22. July 2020 at 04:39

I have been reading The New Yorker for years. I think they have cut back on their vaunted fact checking department as revenues have dropped. This was something that would have been caught at an earlier time.

Regarding Professor Lepore, I really enjoyed reading her recnet history of the United States. Her podcast ‘The Last Archive’ has been an enjoyable listen.

10. derek
22. July 2020 at 05:55

I’m not sure we can throw the NYer fact-checkers under the bus here. Their specialty is not really numeracy but making sure that quotes are true. Here, they could have followed Lepore’s misreading to the Harvard press release and found a quote that is really just extremely badly written, and context from the Harvard release doesn’t help much either.

It is very disappointing that neither Lepore nor the fact-checkers had a very good idea of what a plausible number could be. That said, I am not seeing Prof Sumner’s path to 4/3 at all.

11. ssumner
22. July 2020 at 08:04

Matthias, I’ve heard, but the reports are so vague I haven’t commented. We don’t need average inflation targeting, we need level targeting.

Jason, You said:

“I dislike Trump immensely but the media will always hold him to account.”

You must be joking. The press was 100 times harder on Bush for Katrina than Trump for Hurricane Maria. Even though Trump’s performance was even worse.

Trump says so many outrageous things, almost every day, that comments the press would hammer other presidents for are barely remarked upon.

Just yesterday he said the census would not count illegal aliens for reapportionment. That’s blatantly unconstitutional, but the GOP could care less about the Constitution. The press mostly ignored the outrage.

foosion, LOL. Nice try, but that’s obviously not what the quotation is saying.

Tacticus, But the real issue is why would the New Yorker publish a fact that was obviously wildly false. You don’t have to go to a study to know that that reported statistic is wrong. I didn’t know the actual number is 0.2%, but I knew it wasn’t 66% without someone telling me. Does no one at the New Yorker have a brain?

Alan, I should say that the whole article is pretty bad, not just the laughable statistical mess-up.

Derek, Both Lepore and the fact checker are to blame. No sane person could read that number and think it’s correct. Isn’t having common sense a requirement these days for being a journalist?

And read the quote again. Both police and auto accidents were 2/3 of ER admissions.

12. Christian List
22. July 2020 at 08:33

It appears to me that whomever wrote the press-release for the study made a terrible error

Tacticus,

Their wording could be better but I don’t think they made a factual error. One can misread press release like Jill Lepore did but only if you don’t think about it at all. I assume/hope she looked into the actual study, or didn’t she?

If the press release said what some people think it says, it would be right in the headline because it’s such an absurdly high number.

I haven’t read her whole essay, but the part in question is quite off-putting. First she gives these completely wrong numbers. Right after that she writes that there are fewer police officers in the US than in Europe. So the number of police officers in the US is not the problem. Then she writes that the real difference is guns. But the wrong numbers she gave were US policemen who allegedly “hit” people. So I assume her wrong number is only about beating. But then how are guns the problem??? There at the latest she should have noticed her grotesque mistake.

And it’s still not corrected! Anyone can make gross mistakes, but at least one must correct them as soon as possible.

13. Tacticus
22. July 2020 at 09:06

@ Prof. Sumner,

Well, the evidence suggests that the people involved in this article have no brain, yes. As soon as I read the quote I was wondering where they had such an insane statistic from, as it was clearly wrong. I have no idea how anyone could possibly accept it as valid and not question it for a second. Especially when, afterwards, 2 seconds of Google-ing told me that, in the US, 300,000 teenagers (16-19) go to the ER every year for driving accidents. I was just curious as to where the error came from. Still not sure where the press release got the wrong statistic from.

@ Christian List,

The press release literally states that ‘Sixty-four percent of the estimated 683,033 injuries logged between 2001-2014 among persons age 15-34 resulted from an officer hitting a civilian.’ That is factually wrong.

I don’t know how someone writing an article like this would not double check the methodology of a study putting forth such a prima facie false statistic – other than that it seems plausible to them, because of their own preconceptions.

Looks like the US population 15-34 is about 87 million, so some 0.06% of them are sent to the ER by the police or security officers per annum. Some back of a napkin maths tell me that that age group had about 43 million ER visits in 2017, with some 55,000, or 0.127%, being due to police or security guards. A long ways from the sixty-four percent or two-thirds.

14. Jason
22. July 2020 at 10:02

🙄🙄
I noticed you didn’t remark on Andrew Cuomo.

About 40 percent to half of Covid deaths were due. To Democratic governors forcing nursing homes to accept COVID positive patients.

15. Jason
22. July 2020 at 10:58

https://nypost.com/2020/06/25/blm-co-founder-describes-herself-as-trained-marxist/amp/

Maybe you haven’t noticed Scott, but there is an organized attempt to destroy the United States from within.

I’d take a stupid mans Richard Nixon, even if he is despicable on a personal level, over Hugo Chavez or Maduro any day.🙄🤦‍♂️

16. Christian List
22. July 2020 at 11:24

@Tacticus

This is what I meant by the bad wording. The sentence is perfectly correct. The press release just does not make it clear once again that the logged data is only about incidents that the researchers have described as “non-fatal legal intervention injuries”.

But this becomes clear from the press release alone, you just have to read everything, which nobody does because the language used is so boring.

Look, they only logged data of non-fatal injuries following “legal intervention” – defined as an injury/wound inflicted by police officers or private security guards.

And in the 14-year period studied, the majority of those logged injuries – 64% – were categorized as “struck by/against”, and stemmed from a physical interaction with officers.

Non-fatal firearm injuries caused just 1% of the estimated 683,033 injuries.

As I said, I think the funniest part is when Jill Lepore writes right afterwards that guns are the real problem. — YES, this might be even true. But then why in the world is she using this study and why does she use it so wrongly? It’s kind of ironic.

17. Tacticus
22. July 2020 at 11:42

@ Christian List,

Okay, can you please explain, in detail, how the sentence is correct?

It states, ‘Sixty-four percent of the estimated 683,033 injuries logged between 2001-2014 among persons age 15-34 resulted from an officer hitting a civilian.’

The study finds that ‘For the period 2001–2014, 683,033 […]
legal intervention injuries treated in EDs occurred for US persons age 15–34.’

Those two sentences are not compatible, in my mind?

18. Christian List
22. July 2020 at 12:09

@Tacticus

I wrote it already but okay. Maybe less detail helps? They registered 683,033 non-fatal injuries following “legal intervention”, they call this “non-fatal legal intervention injuries”.

And from those 683,033 cases, 64% were police officers (or private security guards) hitting a civilian. 1% was shooting. And so on.

The percentage is more about WHAT those police officers (or private security guards) did exactly, the “who” has already been settled.

19. Zephito
22. July 2020 at 12:18

This is what it shows now; I presume it was changed though I never saw the original:

“One study suggests that, among American men between the ages of fifteen and thirty-four, the number who were treated in emergency rooms as a result of injuries inflicted by police and security guards was almost as great as the number who, as pedestrians, were injured by motor vehicles.”

I don’t know if this is accurate or not but it sounds more plausible than “66% of young adult emergency room visits are caused by police beatings.”

20. Tacticus
22. July 2020 at 13:55

@ Christian List,

Where are you getting this statistic that 64% of the 683,033 estimated cases were situations in which police officers or private security guards hit civilians?? As opposed to shootings or car violence etc.

I’m not saying you’re wrong, but I am very confused where you’re getting these data from.

21. Christian List
22. July 2020 at 14:47

@Tacticus

The shooting percentage is from the Guardian article mentioned in the press release.

https://www.theguardian.com/us-news/2016/sep/09/police-involved-injuries-civilian-harvard-study

But you don’t need the shooting percentage or the Guardian article to understand the press release. I agree that the language is kind of dry and clumsy though.

@Zephito
You are right, they changed it now. The original wording is quoted right at the beginning of Scott’s blog post.

22. ssumner
22. July 2020 at 16:05

Jason, Not surprised that someone that he thinks BLM is trying to destroy America and that 40% of deaths were due to Democratic governors would also not realize he was commenting on the wrong post. This post is on the New Yorker.

23. Tacticus
23. July 2020 at 03:30

@ Christian List,

Ahh, okay, thanks for clearing that up. Silly me, thinking the press release was about the study, not ancillary information.

24. Christian List
23. July 2020 at 04:54

@Tacticus
Sorry, I can’t agree with you on this one. You even found the two essential sentences of the press release. The wording could be a bit better and they might be too far away from each other, but that’s about it. The relevant information is in those two sentences.

25. Tacticus
23. July 2020 at 05:36

Yep, the sentences are completely accurate, I just made an erroneous assumption 🙂

26. derek
23. July 2020 at 07:28

Scott, in my view, the quote is clearly parsed as: 2/3 of Americans 15-34 in emergency rooms were there for injuries caused by police, which is XXX many people; for ALL Americans of any age, there were approximately the same number XXX injuries of pedestrians by cars.

The quoted sentence is reasonably unambiguous, with the key phrase being “as many” instead of “about the same proportion”. I don’t think your own misinterpretation is doing much to help your case that the NYer folks’ misreading was so insane!

27. ssumner
23. July 2020 at 08:23

Derek, When the first part of a sentence is talking about proportions, and when the comparison does not suggest otherwise, then the reader is naturally going to assume that the second section is also about proportions.

Your interpretation might be right, but in that case you would be arguing that the writer is even stupider than I assumed, as your interpretation is that they are making a completely nonsensical apples and oranges comparison. I prefer to be a bit more generous, and assume that a New Yorker journalist knows how to at least write properly, even if they are innumerate.

[I’m the opposite. Not a good writer, but at least not completely innumerate.]

You said:

LOL. I’m not accusing them of “misreading”, I’m accusing them of being stupid. If you see something that is obviously wrong, don’t put it in your article without checking it. Even if the article had actually said 2/3rds, the New Yorker never should have reported an obviously false fact.

28. Christian List
23. July 2020 at 09:07

derek is right of course, that’s how she meant it. But Scott is right in the sense that the comparison she is making is really absurd.

She is comparing the number of people aged 15-34 who were beaten by police with the number of pedestrians hit by cars, and this is suppossed to tell us what exactly???

Should one find it reassuring or disturbing that the quality of some of these texts is so moderate? I mean, this is pretty much the absolute elite writing in one of the best mediums there is.

Right now I find it reassuring that this kind of elite is not as intellectually advanced as one might think at first glance.

29. Tacticus
23. July 2020 at 09:24

I think Derek is right about what she meant and you, Prof. Sumner, are right that she is even stupider than you initially imagined and that most people would assume the second phrase is also about proportions.

The New Yorker did misread the study, they were then also stupid enough to think that their misreading was correct.

30. ssumner
24. July 2020 at 10:00

I try to be generous in my interpretation of the writing of others.